Teachers Making Sense of Algorithms and Their Politics: Design Reflections From a Professional Development Institute

• In Event: 43.052 - Politics in the Foreground and Background of Learning: Analyses of Power and Learning Across Settings

Abstract

Algorithms have been commonly viewed as confined to the fields of mathematics and computer science (CS). Yet in recent months, algorithms, defined as a chain of commands computers execute to perform a task, have gained a higher profile in academic, activist, and public discourses. Whether we consider allegations of Facebook’s involvement in the 2016 US presidential elections (Allcott & Gentzkow, 2017), the rebellion at Google over the company’s involvement with the Pentagon (Conger, 2018), or Microsoft employees denouncing the use of their deep learning algorithms by Immigration and Custom Enforcement (Lecher, 2018), we are in a moment where the politics of algorithms are being researched, exposed, and passionately debated (Eubanks, 2018; Nobel, 2018; O’Neill, 2016). Indeed, new technologies and the algorithms driving them are profoundly, albeit often silently, shaping the world around us. Yet, teaching and learning about the politics of algorithms is not an educational priority in our nation’s schools. This is the case despite the current movement to expand and formalize CS learning in K-12 schools, which is primarily fueled by economic or national competitiveness rather than student or community empowerment (Vakil, 2017). There are scarce opportunities for students (or teachers) to learn how algorithms are intertwined with systems of power.

In this study, we sought to address this gap through the design of a week-long professional development institute for non-STEM preservice teachers at Shellbrook, a historically Black college in the South. Our design is motivated by the belief that learning about the ethics and politics of technology is a crucial and often overlooked aspect of a liberatory educational agenda (Freire, 1997) as well as a belief that teaching and learning about (the politics of) algorithms should not be confined to CS or even STEM students. Activities illuminated the influence of human biases on the decisions that are made by algorithms and through machine learning, such as using human generated datasets that aid self-driving cars in making moral decisions (Bonnefon, Shariff, & Rahwan, 2016) and how selection of training data can lead to racial and gender bias in facial recognition (Buolamwini, 2017). In the culminating activity, participants designed a classroom lesson related to algorithms that was appropriate for their grade and content area. Our dataset is comprised of audio/video recordings, student workbooks, online surveys, and both physical and digital artifacts created by participants.

Our analysis focuses on investigating how specific features of the pedagogical design created opportunities for sensemaking about the politics of algorithms. Preliminary findings suggest that exposing non-STEM preservice teachers to various ways that algorithms impact their life provided an opportunity for the demystification of algorithms. In line with our design principles, we found it was important to provide multiple opportunities to deconstruct algorithms before exploring how bias and power shaped algorithmic processes. In addition, focusing on personal connections made the topic of algorithms accessible to non-STEM preservice teachers. Finally, providing an opportunity to apply their understanding of algorithms to their content area helped inform ways in which we can improve the PD in future iterations.

Authors

• Sepehr Vakil, Northwestern University

• Jason Harron, The University of Texas at Austin

• Gareth Gingell, The University of Texas at Austin

• Tatiane Russo-Tait, The University of Texas at Austin